Gender Norm Salience across Middle Schools: Contextual Variations in Associations between Gender Typicality and Socioemotional Distress

Abstract

Youth who feel they do not fit with gender norms frequently experience peer victimization and socioemotional distress. To gauge differences between schools, the current study examined the longitudinal effects of school-level gender norm salience—a within-school association between gender typicality and peer victimization—on socioemotional distress across 26 ethnically diverse middle schools (n_boys = 2607; n_girls = 2805). Boys (but not girls) reporting lower gender typicality experienced more loneliness and social anxiety in schools with more salient gender norms, even when accounting for both individual and school level victimization. Greater gender norm salience also predicted increased depressed mood among boys regardless of gender typicality. These findings suggest particular sensitivity among boys to environments in which low gender typicality is sanctioned.

Introduction

Gender typicality is one dimension of gender identity (Egan & Perry, 2001), reflecting a sense of fit with, or conformity to, gender norms. Not feeling like a “typical” boy or a girl is distressing for children (Carver, Yunger, & Perry, 2003) and adolescents (Menon, 2011), and in early adolescence fitting in is highly important (Blakemore & Mills, 2014). Experiences with bullying or peer victimization account for at least part of the socioemotional distress associated with low gender typicality (e.g., D. S. Smith & Juvonen, 2017). Despite bullying frequently taking place in school, however, surprisingly little is known about how strongly peer victimization is related to gender atypicality across schools. It is conceivable that students who are less gender typical are at greater risk of peer mistreatment in some schools compared to others, which may in turn make gender norms more salient. As such, it is critical to understand differences in the salience of gender norms between schools when considering socioemotional distress of adolescents who do not feel gender typical. In schools where peers frequently taunt and ridicule students who deviate from gender norms, gender norms become highly salient and the experience of feeling gender atypical may be especially stressful. Moreover, attending a school where peers target those who are less gender typical may increase the distress of all youth, inasmuch as it can increase gender conformity pressures.

Gender Norms and Norm Salience

Social groups expect and value particular sets of behaviors and traits in their members. Individuals who adhere to these norms tend to be socially rewarded (i.e., have higher social status; Stormshak, Bierman, Bruschi, Dodge, & Coie, 1999), while those who deviate from social norms are likely to be “punished” (e.g., rejected, bullied) by members of the collective (Wright, Giammarino, & Parad, 1986). Regarding gender, specifically, those who do not conform to norms or meet expectations of what it means to be a typical boy or girl are more likely to be bullied (Pascoe, 2012) or excluded by peers (Heinze & Horn, 2014). Those reporting low gender typicality are at an increased risk for socioemotional problems during childhood (Carver et al., 2003) and adolescence (Menon, 2011), and peer mistreatment can account for at least part of this distress. Concurrently, both gender-based teasing (Jewell & Brown, 2014) and lack of acceptance (T. E. Smith & Leaper, 2006) have been shown to account for associations between low gender typicality and worse socioemotional outcomes during adolescence, and a similar pattern has been shown over time (D. S. Smith & Juvonen, 2017).

To understand the socioemotional distress of youth who do not feel gender typical, the current study is designed to shed light not on social experiences of individual youth, but on differences between schools, by focusing on gender norm salience. Rather than focusing on student characteristics or norms (e.g., aggression) that are made salient through their association with high status (e.g., popularity; Dijkstra & Gest, 2015), we propose that norms become salient also when youth learn what is not accepted at their school by observing what is not tolerated by their peers (cf. Schwalbe & Mason-Schrock, 1996). For example, for a boy who is ridiculed for not being masculine enough, or a girl about whom rumors of sexual promiscuity are spread, gender norms surrounding masulinity and sexuality, respecitvely, are likely to stand out. Attending to differences between schools, specifically in terms of whether low gnder typicality is related to an increased risk of peer ridicule and intimidation, provides important insights not only about salience of gender norms, but also about socioemotional distress in early adolescence. We expect that gender norm salience—assessed at a collective level—functions as a contextual moderator for youth who do not feel gender typical. That is, at schools in which gender norms are less salient, gender typicality might not be associated with socioemotional distress. By contrast, those who do not perceive themselves to be “like the typical boy or girl” may experience intensified socioemotional distress in environments where gender norms are more salient (i.e., where norm violations are negatively sanctioned). The question is whether such school differences can account for the socioemotional distress of adolescents—particularly those who feel less gender typical—over and above their personal experiences of peer victimization.

Increased salience of gender norms (i.e., contexts where low gender typicality increases risk for peer victimization) may also negatively affect the socioemotional distress of all students, not just those who are less gender typical, by increasing gender conformity pressures in both implicit and explicit ways (cf. Juvonen & Galvan, 2009). For example, research on peer victimization that targets body weight shows that in schools with higher rates of such peer mistreatment, all students—regardless of their weight—report greater internalizing problems (Lampard, MacLehose, Eisenberg, Neumark-Sztainer, & Davison, 2014). The authors presume that these findings reflect the effects of the negative climate of the school. Other research demonstrates that perceived pressure to conform to gender norms is associated with greater socioemotional distress among children and young adolescents, regardless of gender typicality (Yunger, Carver, & Perry, 2004). Such findings are robust across children, adolescents (Egan & Perry, 2001), and adult men (e.g., Good, Robertson, Fitzgerald, Stevens, & Bartels, 1996). Hence, if stronger negative associations between gender typicality and victimization impart peer pressure to conform to gender norms (e.g., to avoid mistreatment by peers), even those who feel that they are gender typical may experience greater socioemotional distress in schools with higher gender norm salience.

Early Adolescence

Gender socialization, such as positive or negative feedback on adherence to gender norms, comes from a variety of sources throughout development, including parents during early childhood (Kane, 2006) and adolescence (Carr, 2007), teachers (Kosciw, Greytak, Palmer, & Boesen, 2014), siblings (e.g., McHale, Crouter, & Tucker, 1999), and peers. Although multiple individuals continue to influence youth’s understanding of gender, peers are frequently an especially strong socializing force during adolescence for several reasons. Although peers begin to monitor each other’s adherence to gender norms early in childhood (e.g., Blakemore, 2003), gender typicality and the implications of “standing out” from peers for socioemotional distress are of particular interest during early adolescence. Following the transition from elementary to middle school, youth experience higher levels of peer victimization at school overall (Pellegrini & Long, 2002). During this same developmental period, sensitivity to peer feedback is heightened, as evidenced, for example, by increased responsivity to exclusion (Sebastian, Viding, Williams, & Blakemore, 2010) and more reported fear of social evaluation (Westenberg, Drewes, Goedhart, Siebelink, & Treffers, 2004), while pubertal development increases the physical salience of gender (Galambos, Almeida, & Petersen, 1990). As such, both peer social feedback on gender norms and the potential impact of this feedback are likely to be heightened during this time.

Due to the nature and pervasiveness of gender differences in what is likely not to be tolerated by peers during adolescence (e.g., Eder, Evans, & Parker, 1995), the socioemotional impact of gender norm salience may differ for boys and girls. Boys tend to face greater conformity pressure throughout development (e.g., Kane, 2006), and gender-related victimization of boys tends to be particularly targeted and salient during middle school (Eder et al., 1995) and high school (Pascoe, 2012). For example, verbal harassment with an emphasis on homophobic epithets (e.g., gay, girly, wimpy; Pascoe, 2012), underscores the punitive reinforcement of heterosexuality and toughness. Girls who are not gender typical are also at risk for being victimized, but are targeted in less uniform ways across development. During childhood in particular, girls may even be encouraged by peers or adults to engage in what are seen as gender atypical behaviors (e.g., be tougher; Kane, 2006). Thus, just as individual-level gender conformity pressures appear to differ by gender, with boys experiencing more uniform and consistent pressure than girls, school levels of gender norm salience and related effects may also vary for boys and girls.

Current Study

Consistent with the view that reactions to norm violations are particularly powerful ways to learn about what is not tolerated or accepted within a group or a collective (Baumeister, Zhang, & Vohs, 2004), we focus in this study on the negative social sanctions associated with low gender typicality. Defining gender norm salience (cf. Cialdini, Kallgren, & Reno, 1991) as the risk of perceived peer victimization among youth who do not feel like typical boys or girls, we examine the ways in which school-specific gender norm salience is related to the socioemotional distress of young adolescents. This approach is novel inasmuch as our goal is to extend past research that examines school gender climate by relying on aggregated self-reports on safety or victimization of those who deviate from gender norms (e.g., Toomey, McGuire, & Russell, 2012). It thus allows for examination of a school-wide pattern without requiring participating youth to have full insight into the causes of their own or others’ victimization experiences. Additionally, by focusing on student attributes (gender typicality) that contribute to negative peer treatment (victimization), we extend past work examining school or classroom norm salience as a function of student behaviors that contribute to high peer status (e.g., popularity; Dijkstra & Gest, 2015). Specifically, focusing on differences between schools, we computed within-school regression slopes (seventh grade peer victimization regressed on seventh grade gender typicality; see Figure 1) to capture the salience of gender norms. Although there are numerous reasons that adolescents may experience victimization, we presume that this school-level association between victimization and gender typicality provides an estimate of their grademates’ negative reactivity to deviations from gender norms, and thus the salience of gender norms in general.

Figure 1.

Conceptual diagram of associations between gender norm salience at the school level (operationalized as peer victimization regressed on gender typicality within schools; top half of figure), gender typicality, and socioemotional distress.

We hypothesized that gender norm salience (presumed to capture low tolerance of deviations from gender norms) moderates the association between gender typicality and socioemotional distress for individual students within the school (see Figure 1). That is, low gender typicality was expected to predict higher socioemotional distress in schools in which gender norms are more salient because low gender typicality is negatively sanctioned. We also tested whether gender norm salience is directly related to socioemotional distress of all students within the school (i.e., regardless of their gender typicality). It was presumed that gender norm salience may also capture greater gender conformity pressures and therefore be directly associated with greater distress for all students.

The regression coefficients representing gender norm salience were computed separately for boys and girls in each school, and analyses were conducted separately for boys and girls. The separate analyses were warranted for two main reasons. First, gender typicality (and thus salience of norms) was measured in a manner that was gender-specific (i.e., measures refer to either other boys or other girls). Second, given gender differences in experiences and content of gender socialization and gender-related victimization, the salience of gender norms is likely to vary between boys and girls within any one school. In other words, a school environment may be high in gender norm salience for boys but not girls, or the opposite may be true. Although we did not analytically compare boys and girls for the aforementioned reasons, we expected that high salience of gender norms would have a particularly strong impact among boys. In addition to our contextual moderator hypotheses, it was expected that in schools with higher salience of gender norms among boys, even gender typical boys may experience distress due to more explicit gender conformity pressures.

To examine the associations between within-school gender norms salience and socioemotional distress, we specifically assessed loneliness, social anxiety, and depressed mood. In the current study, gender typicality and peer victimization relied on self-reports, and as such, we capture subjective realities that are highly relevant when considering socioemotional outcomes (e.g., Juvonen, Nishina, & Graham, 2001). Gender norm salience is assessed separately for boys and girls within each school during the second year of middle school (i.e., seventh grade when social norms are likely to have been established) and we examine its effects on socioemotional distress during the last year of middle school (i.e., eighth grade), controlling for baseline distress during the first year (i.e., sixth grade).

Method

Participants

Participants were part of a longitudinal study of 5,991 early adolescents (52% girls) recruited from 26 public middle schools. The sample was ethnically diverse, (31% Latino/a, 20% White, 13% East/Southeast Asian, 12% Black/African American, 14% multiethnic/biracial, 9% other, and 1% unreported), with specific ethnic composition varying by school. The number of participating students in each school ranged from 78 to 445 (M=281.57, SD= 111.68). The current analysis sample of 5,412 participants (52% girls) included those who identified as part of one of the largest panethnic groups (35% Latino/a, 22% White, 15% Asian, 13% Black) or as multiethnic (15%). Data from youth reporting “other” ethnic backgrounds were omitted from the initial sample of 5,991, as this was a very heterogeneous group, making it problematic in terms of controlling for ethnic differences in socioemotional distress. Of the participants in our analytic sample, 98% participated in fall of sixth grade, 96% in spring of sixth grade, 86% in spring of seventh grade, and 78% in spring of eighth grade. The participation rates in later waves are comparable to other longitudinal studies with largely ethnic minority samples in urban school settings (e.g., Seidman, Allen, Aber, Mitchell, & Feinman, 1994).

Because analyses were conducted separately for boys and girls, gender-specific comparisons of participants whose data were used and those who were excluded from the analyses (because they reported “other” or no ethnicity) were conducted. Boys who were not included in the current analyses reported similar levels of gender typicality, peer victimization, and baseline depressed mood as boys who were included. Boys who were not included reported slightly higher baseline social anxiety (M= 2.15, SD = .86; t(2685) = 3.28, p = .001) and loneliness (M = 1.88, SD = .95; t(1779) = 2.68, p = .008) compared to their peers who were included (social anxiety: M= 1.98, SD = .76; loneliness: M= 1.70, SD = .82). Girls who were not included in the analyses reported similar levels of gender typicality, victimization, and baseline depressed mood and loneliness as those who were included. Girls who were not included reported slightly higher baseline social anxiety (M= 2.25, SD = .84; t(2866) = 2.13, p = .033) than their peers who were included (M= 2.14, SD = .77).

Procedure

Students were recruited during fall of their sixth grade year from 26 urban middle schools in California. Due to the language demands of the survey, students in self-contained English Language Learner or Special Education programs were not recruited for the study. The current study relied on data collected during sixth, seventh, and eighth grades. Participating students received written consent from a parent or guardian, and consent forms were provided in languages other than English (Spanish, Chinese, Vietnamese, and Korean) as necessary. To encourage students to return the forms, two $50 gift cards were raffled at each participating school for students who returned signed consent forms, whether or not they were given permission to participate. In addition, two iPods were raffled to study participants. Overall, 81% of the consent forms distributed were returned, with 83% of these students receiving consent to participate. Participants provided written assent before completing surveys, then completed written questionnaires in the classroom setting during fall and spring of sixth grade, spring of seventh grade, and spring of eighth grade. Instructions and questions were read aloud by a trained researcher. Students received an honorarium for taking each survey ($5 for each sixth grade survey and $10 for seventh and eighth grade surveys). The survey and all procedures were approved by the University Institutional Review Board.

Measures

Demographic data.

Participants indicated whether they were male or female during sixth grade, and this indication of biological sex was used as a proxy for gender. Participants reported their ethnicity in sixth grade, using a list of 13 options including a write-in option. These were further collapsed into the aforementioned panethnic categories. When providing consent, parents reported their highest level of education, using categories ranging from elementary/junior high school (1) to graduate degree (6). Preliminary analyses also included an index of school-level ethnic diversity (shown to be associated with social anxiety and loneliness; e.g., Juvonen, Nishina, & Graham, 2006). This index (Simpson, 1949), ranging from 0 to 1, indicates the probability that any two children chosen at random at a school will be of the same ethnicity (with higher values indicating greater diversity).

Gender typicality.

Participants’ self-perceptions of gender typicality were assessed during seventh grade using three items adapted from a six-item measure (Egan & Perry, 2001). Items on the boys’ surveys included “I feel I am just like other boys,” “I feel I fit in with other boys,” and “I feel I act just like other boys,” with respective rewording for girls. Participants indicated their responses on a five-point scale ranging from “Not at all” (1) to “All the time” (5). The three items had good internal consistency for boys and for girls (α = .74 for both) and were averaged with higher values indicating a stronger sense of gender typicality. These scores were used to compute the regression slopes that define gender norm salience in each school (see below), and they also function to capture individual-level gender typicality. All items refer specifically to a sense of fitting into one’s gender category and will be discussed in terms of higher and lower gender typicality (rather than typicality and atypicality).

Peer victimization.

To assess peer victimization during seventh grade, we relied on four items from a self-reported peer victimization measure designed to reduce self-report bias (Neary & Joseph, 1994). Participants first chose between two statements (e.g., “Some kids are often picked on by other kids” or “Other kids are not picked on by other kids”). They then indicated their level of agreement with that statement (“Really True for Me” or “Sort of True for Me”), creating a four-point scale for each item. The four items were internally consistent (α = .77) and were averaged such that higher scores indicated greater peer victimization for each participant. The scores were also aggregated within schools to capture the average peer victimization as a school-level control. Additionally, peer victimization scores were used to compute school-level gender norm salience slopes, as shown below.

Within-school gender norm salience.

The association between gender typicality and victimization (measures described above) was computed separately for boys and girls at each school site to capture gender norm salience at seventh grade. Our norm salience measure captures how strongly feelings of gender typicality (or lack thereof) were related to perceptions of getting victimized by peers. To compute gender norm salience within each school, ordinary least squares (OLS) estimation was used to estimate the regression slope of peer victimization on gender typicality. This slope was estimated separately for boys and girls within each school because the social reference group for these questions differed by gender (i.e., boys were asked specifically about typicality in relation to other boys, and girls in relation to other girls).

The multilevel literature suggests that there is no single best way to estimate random slopes such as those used to define gender norm salience; OLS estimates are unbiased but noisy, whereas empirical Bayesian estimates are biased but exhibit less variability (Raudenbush & Bryk, 2002). We opted for the former approach based on its unbiasedness and because we did not want to risk underestimating this important source of school-level variation as might be the case with empirical Bayesian estimates. To improve interpretability of the effects, we scaled gender norm salience such that higher, positive values corresponded to a stronger negative association between gender typicality and peer victimization. Thus, the within-school OLS slopes were multiplied by −1. The gender norm salience values were standardized across schools, separately for boys and girls. As such, a value of 0 for boys’ or girls’ gender norm salience indicated a school with an average level of gender norm salience for that gender category, whereas a value of 1 indicated that a school’s level of gender norm salience was 1 standard deviation above average (i.e., lower gender typicality more strongly related to higher peer victimization).

Socioemotional distress.

Participants’ socioemotional distress was measured twice: during their first and last year in middle school (i.e., sixth and eighth grades) to assess baseline and outcome distress, respectively. Baselines for social anxiety and loneliness were measured during the fall of sixth grade; loneliness was not measured during the first wave of data collection, and as such the baseline used was from spring of sixth grade. Loneliness was measured using 5 items (α₆ = .91, α₈ = .92) adapted from an established measure (Asher & Wheeler, 1985). Participants were asked about how they felt at school using statements such as “I have nobody to talk to” and “I feel left out of things” rated on a five-point scale (1-Always true, to 5-Not true at all). Social anxiety was measured using six items (α₆ = .78, α₈ = .81) from the Social Anxiety Scale for Children-Revised (LaGreca & Lopez, 1998). These items were from two subscales: Fear of Negative Evaluation (e.g., “I worry about what others think of me”) and Social Avoidance and Distress—General (“It’s hard for me to ask others to do things with me”). Participants rated how true these statements were for them on a five-point scale (1-Not at all, to 5-All the time). Depressed mood was measured using seven items (α₆ = .80, α₈ = .85) adapted from the Center for Epidemiological Studies Depression Scale (CES-D; Radloff, 1977). Participants indicated how often in the past week they had experienced symptoms (e.g., “I felt sad”) on a four-point scale (1-“less than 1 day” to 4-“5-7 days”). All measures of socioemotional distress captured the means of respective items, and higher values indicated more distress.

Analytical Plan

As previously mentioned, all analyses were conducted separately for boys and girls because (a) gender typicality was measured using items specifically referencing “other boys” or “other girls” and (b) gender-related victimization is qualitatively different for boys and girls, as previously described. Gender typicality was also grand-mean centered separately for boys and girls. To test our main hypotheses, multilevel regression models (students nested within schools) with random intercepts were used to examine the effects of seventh grade individual-level gender typicality (GT_ij), seventh grade school-level gender norm salience (GNS_j), and the cross-level interaction of gender typicality and gender norm salience (GT_ijGNS_j) on socioemotional distress (SD_ij). All analyses also controlled for student ethnicity (as the largest ethnic group, Latino students served as the reference group), grand-mean centered baseline (sixth grade) socioemotional distress and seventh grade peer victimization, and grand-mean centered school-level peer victimization (COV_k), capturing overall level of peer mistreatment. The equation for the main analysis model is presented below.

$$S{D}_{ij}={\text{γ}}_0+{\text{γ}}_1(G{T}_{ij})+{\text{γ}}_2(GN{S}_j)+{\text{γ}}_3(G{T}_{ij})(GN{S}_j)+{\displaystyle \sum _{k=1}^k{\text{γ}}_{k+3}}(CO{V}_k)+u_{0j}+{\epsilon }_{ij}$$

Preliminary models also included parent education and school-level ethnic diversity; as these did not predict distress, they were excluded from final models. For models in which the interaction between gender typicality and gender norm salience was not significant, models were re-run without this interaction in order to correctly interpret the effects of individual-level gender typicality and school-level gender norm salience.

In the current study, gender typicality, baseline loneliness, and all distress outcomes at eighth grade were part of a planned missing design in order to maximize efficiency of data collection while reducing the burden on participants (Graham, Taylor, Olchowski, & Cumsille, 2006). For each of these measures, only two thirds of the participants (randomly selected) were asked to respond; these data were thus assumed to be missing completely at random (MCAR). Additional unplanned missingness did not exceed approximately 20%. Although there is no way to confirm that the unplanned missing data are missing at random (MAR), there was also no reason to expect that missingness on these variables was related to the constructs themselves. Rather, attrition in later waves due to school transfers and student absence likely accounts for a majority of unplanned missingness. Because we presume that missing data were MCAR or MAR, analyses were conducted in MPlus (Muthen & Muthen, 1998-2016) using full information maximum likelihood (FIML) with robust standard errors. FIML is considered an appropriate method to account for data that is assumed to be MAR or MCAR, and is considered preferable when data is missing at rates higher than 5-10%, as techniques such as listwise deletion can bias results (Little, Jorgensen, Lang, & Moore, 2014).

Results

The results are reported in two sections. First, we examined descriptive statistics for school- and individual-level variables. Intraclass correlations (ICCs) were computed to assess the proportion of variance accounted for across schools, and are included as well. In the second section, we report the results of our main analyses: multilevel models across the three outcomes for boys followed by girls.

Descriptive Statistics

Figure 2 shows unstandardized OLS regression coefficients depicting school-level gender norm salience of boys and girls. Because slope values were multiplied by −1, higher (positive) gender norm salience values correspond to a stronger negative association between gender typicality and victimization (i.e., students who are less gender typical are more victimized), whereas negative slopes correspond to a positive association between gender typicality and victimization (i.e., students who are more gender typical are more victimized). As expected, gender norm salience varied across the 26 schools for boys and girls. For girls, unstandardized OLS coefficients ranged from B = .01 to .34 (M =.18, SD = .07). For boys, the school-level gender norm salience values ranged from B = −.12 to .58 (M= .13, SD = .11) across the schools. As expected, these values were all positive with one exception; for boys in one school, the coefficient was −.12 (n.s.; see Figure 2). Gender norm salience values for boys and girls across the school sites were not significantly correlated (r = .27, p = .177), suggesting that there is not a strong association between gender norm salience of boys and of girls, and supporting the decision to examine gender norm salience of boys and of girls separately.

Figure 2.

Distribution of unstandardized gender norm salience coefficients for boys and girls across schools.

ICCs were computed to determine the proportion of between-school variance in each of the outcomes. The ICCs for loneliness (girls = .04, boys = .04), social anxiety (girls = .02, boys = .06), and depressed mood (girls = .01, boys = .01) were small, meaning that most of the variability was at the individual level, rather than between schools. Although the variability between schools was relatively small, descriptive statistics, as well as final models, took nesting of individuals within schools into account.

Means and standard deviations for individual level variables are provided in Table 1. To examine differences between boys’ and girls’ average levels of gender typicality and each outcome while accounting for the nested nature of the data, we tested a series of random intercept multilevel models with gender as the sole predictor. Compared to boys, girls reported lower gender typicality (B = −.27, S.E. = .03, p < .001) and victimization (B = −.08, SE= .02, p = .002), and higher levels of loneliness (B = .23, S.E. = .03, p < .001), social anxiety (B = .23, S.E. = .03, p < .001), and depressed mood (B = .35, S.E. = .03, p < .001). Correlations for individual seventh grade gender typicality and peer victimization, sixth grade baselines for emotional distress, and eighth grade socioemotional distress outcomes are shown in Table 2; variables were centered within schools (i.e., group-mean centered) prior to calculating correlations in order to obtain estimates of within-school associations. Gender typicality, peer victimization, baseline socioemotional distress, and socioemotional outcomes were significantly correlated for both boys and girls. Participants reporting lower gender typicality in seventh grade also reported higher levels of peer victimization in seventh grade and socioemotional distress in both 6^th and 8^th grade. Likewise, those reporting higher levels of peer victimization also reported higher socioemotional distress.

Table 1.

Means and Standard Deviations for Individual-Level Variables.

	Boys		Girls
	Mean	SD	Mean	SD
Gender Typicality (7th grade)	3.56	.94	3.28	.95
Peer Victimization (7th grade)	2.02	.78	1.95	.79
Baseline Loneliness (6th grade)	1.67	.78	1.62	.76
Loneliness (8th grade)	1.59	.78	1.82	.87
Baseline Social Anxiety (6th grade)	1.98	.76	2.14	.77
Social Anxiety (8th grade)	1.97	.75	2.20	.76
Baseline Depressed Mood (6th grade)	1.54	.54	1.62	.58
Depressed Mood (8th grade)	1.52	.54	1.87	.71

Table 2.

Correlations Between Primary Individual-Level Gender Typicality, Peer Victimization, and Socioemotional Distress.

	Gen. Typ.	Peer Vic.	Lon. (6th)	Lon. (8th)	Soc. Anx. (6th)	Soc. Anx. (8th)	Dep. Mood (6th)	Dep. Mood (8th)
Gender Typicality  (7th grade)	---	−.21***	−.192***	−.24***	−.15***	−.18***	−.17***	−.19***
Peer Victimization  (7th grade)	−.16***	---	.15***	.15***	.21***	.16***	.22***	.23***
Baseline Loneliness  (6th grade)	−.21***	.20***	---	.35***	.38***	.30***	.29***	.18***
Loneliness(8th   grade)	−.18***	.15***	.45***	---	.24***	.51***	.25***	.48***
Baseline Social  Anxiety (6th grade)	−.15***	.18***	.40***	.23***	---	.36***	.29***	.19***
Social Anxiety (8th   grade)	−.13***	.17***	.38***	.55***	.37***	---	.23***	.39***
Baseline Depressed   Mood (6th grade)	−.16***	.21***	.28***	.21***	.43***	.21***	---	.35***
Depressed Mood   (8th grade)	−.09**	.17***	.27***	.46***	.19***	.43***	.33***	---

Note. Values for boys are below the diagonal and values for girls are above the diagonal. All variables were group-mean centered to reflect within-school associations between the individual level variables.

^**p<.01.

^***p < .001.

Multilevel Models

Boys.

Coefficients and standard errors from multilevel models for boys are shown in Table 3. Compared to Latinos, all other panethnic groups reported greater feelings of loneliness. Asian and White boys reported higher levels of social anxiety than Latinos, and Asian boys also reported higher levels of depressed mood. Higher baseline socioemotional distress, as well as higher individual-level victimization, predicted greater eighth grade distress across all three outcomes, as expected.

Table 3.

Coefficient Estimates and Standard Errors for Predictors of 8^th Grade Socioemotional Distress Outcomes for Boys

	Loneliness		Social Anxiety		Depressed Mood
	B	SE	B	SE	B	SE
Individual Level
Intercept	1.45***	.04	1.92***	.04	1.49***	.02
Black	.13*	.07	.02	.05	−.05	.06
Asian	.37***	.08	.36***	.06	.15***	.04
White	.23***	.07	.16**	.06	.07	.04
Multiethnic	.23**	.07	.07	.05	.07	.05
Victimization	.08***	.03	.10***	.02	.08***	.02
Baseline Adjustment	.40***	.04	.33***	.03	.31***	.03
Gender Typicality	−.07**	.02	−.05*	.02	−.01	.03
School Level
Victimization	−.06	.13	−.04	.14	−.08	.08
Gender Norm Salience	.04*	.02	.02	.02	.04***	.01
Cross Level
Gender Typicality X Gender Norm Salience	−.04*	.02	−.04***	.01	.01	.02

Note. Bolded rows highlight key predictors for our hypotheses.

^*p<.05.

^**p<.01.

^***p < .001.

Controlling for ethnic differences, baseline distress, and individual- and school-level peer victimization, the associations between gender typicality and both loneliness and social anxiety were moderated by gender norm salience at the school level. Tests of simple slopes were conducted comparing socioemotional distress of boys with varying levels of gender typicality at schools with low (one standard deviation below average), average, and high (one standard deviation above average) gender norm salience. Note that most discussion of gender norm salience is in relative terms (e.g., more vs. less); the phrases “low gender norm salience” and “high gender norm salience,” in turn, are used to refer specifically to schools one standard deviation below or above average in the context of examining simple slopes. We also decomposed the interaction to examine the effects of school-level gender norm salience among students at varying levels of gender typicality.

Consistent with our moderation hypothesis, as gender norm salience increased, the negative association between gender typicality and loneliness became stronger (see Figure 3a). Specifically, whereas gender typicality was unrelated to loneliness for boys in schools with low gender norm salience (B = −.03, SE = .03, p =.274), gender typicality was negatively related to loneliness in schools with average (B = −.07, SE = .02, p = .004) and high gender norm salience (B = −.11, SE = .03, p = .001). That is, feeling less like other boys was related to feeling lonelier among boys attending schools with high, but not low, gender norm salience. We also decomposed the interaction to understand the main (conditional) effect of school-level gender norm salience. More salient gender norms predicted an increase in loneliness for boys who perceived themselves to be at an average level of gender typicality. This supports the hypothesis that gender norm salience might affect boys regardless of their gender typicality.

Figure 3a-b.

The effects of 7^th grade gender typicality and gender norm salience on 8^th grade distress among boys. 3a. Interaction between gender typicality and gender norm salience in predicting loneliness. 3b. Interaction between gender typicality and gender norm salience in predicting social anxiety.

The pattern of findings for social anxiety was similar and is also consistent with our moderation hypothesis. As gender typicality decreased (i.e., boys felt that they were less like other boys), social anxiety increased for boys at schools with average (B = −.05, SE = .02, p = .026) and high gender norm salience (B = −.09, SE= .03, p= .002), but not at schools with low gender norm salience for boys (B = .00, SE = .02, p = .846; Figure 3b).

In contrast to loneliness and social anxiety, no significant cross-level interaction was obtained for depressed mood. In order to interpret main effects instead of lower order effects, the interaction term was then excluded from the model. The coefficients for gender typicality and gender norm salience remained similar to the conditional effects reported in Table 3. Individual-level gender typicality was not a significant predictor of depressed mood (B = −.01, SE= .03, p = .702), suggesting that boys who did not feel gender typical were no more depressed than their more typical peers. However, gender norm salience remained a significant predictor of depressed mood (B = .04, SE = .01, p < .001), indicating that boys at schools with more salient gender norms (i.e., a stronger negative association between gender typicality and peer victimization) reported higher levels of depressed mood, regardless of their self-perceptions of gender typicality.

Taken together, findings for boys largely support our hypotheses. Consistent with the contextual moderation hypothesis, in schools with more salient gender norms, boys who viewed themselves as less gender typical fared worse in comparison to their more gender typical counterparts in terms of social anxiety and loneliness by eighth grade. In addition, even boys who saw themselves as very gender typical reported higher levels of depressed mood in schools with more salient gender norms. These school level effects were documented over and above general school-level peer victimization and, most importantly, beyond boys’ personal experiences of peer victimization.

Girls.

Coefficients and standard errors from regression models for girls are shown in Table 4. Individual level covariates revealed that Black girls reported lower levels of loneliness than Latinas, and Asian and multiethnic girls reported higher levels of loneliness. Black girls also reported lower levels of social anxiety than Latinas, and Asian and White girls reported higher levels of social anxiety. As expected, higher baseline socioemotional distress and higher individual-level peer victimization predicted greater eighth grade distress across all three outcomes. Higher average victimization at the school level was unrelated to loneliness or social anxiety, but negatively associated with depressed mood.

Table 4.

Coefficient Estimates and Standard Errors for Predictors of 8^th Grade Socioemotional Distress Outcomes for Girls

	Loneliness		Social Anxiety		Depressed Mood
	B	SE	B	SE	B	SE
Individual Level
Intercept	1.79***	.05	2.12***	.05	1.84***	.04
Black	−.16*	.08	−.15*	.07	−.03	.06
Asian	.21***	.06	.25***	.06	−.01	.06
White	.11	.06	.19**	.06	−.02	.07
Multiethnic	.19**	.07	.10	.06	.12	.07
Victimization	.11***	.04	.10***	.03	.14***	.02
Baseline Adjustment	.33***	.05	.32***	.03	.36***	.04
Gender Typicality	−.18***	.03	−.10***	.02	−.09***	.02
School Level
Victimization	−.03	.11	−.09	.13	−.24*	.10
Gender Norm Salience	.06	.03	.03	.02	.01	.03
Cross Level
Gender Typicality  X Gender Norm Salience	−.01	.02	.02	.02	.01	.01

Note. Bolded rows highlight key predictors for our hypotheses.

^*p<.05.

^**p<.01.

^***p < .001.

As shown in Table 4, controlling for ethnic differences, baseline distress, and individual- and school-level victimization, there was not a significant interaction between gender typicality and gender norm salience for girls for any of the measures of socioemotional distress. In other words, the association between seventh grade gender typicality and eighth grade socioemotional distress did not differ based on salience of gender norms across schools. To interpret the main effects of gender typicality and gender norm salience, the nonsignificant interactions were removed from the model. The coefficients for main effects of gender typicality and gender norm salience remained similar to the conditional effects reported in Table 4. Gender typicality remained a significant predictor of distress among girls (loneliness: B = −.18, SE= .03, p < .001; social anxiety: B = −.10, SE= .02, p < .001; depressed mood: B = −.09, SE = .02, p < .001), while gender norm salience was not associated with distress. That is, controlling for sixth grade socioemotional distress and other covariates, girls who reported lower gender typicality at seventh grade reported higher levels of loneliness, social anxiety, and depressive symptoms by eighth grade, regardless of the salience of gender norms at their school.

Taken together, findings for girls indicate that lower sense of gender typicality was consistently related to greater socioemotional distress among girls regardless of their school environment. Gender norm salience did not moderate any of the associations, nor did it have any main effects on girls’ distress. Although boys’ and girls’ experiences were not compared directly, our findings for the two groups align with the hypothesis that the effects of gender norm salience would be particularly strong for boys rather than for girls.

Discussion

Feeling that one is not like a typical boy or girl is a developmental liability inasmuch as it increases the risk of socioemotional difficulties, particularly in early adolescence. The question is whether there are environments in which the effects of such a liability are intensified—or ameliorated. We posited that identifying schools where less gender typical youth perceive themselves to be more targeted by their peers represents one way to capture school-wide intolerance or lack of inclusivity of gender atypicality. This approach aligns with the idea that public punishment or derogation of norm violations is a powerful way to convey what is not accepted or tolerated within a particular community (Baumeister et al., 2004). Less tolerant school environments were then expected to intensify the socioemotional distress of less gender typical youth. Assuming that gender norm salience might capture an aspect of the school gender climate, we additionally examined whether gender norm salience would be related to increased distress of gender typical youth as well. As far as we know, the current investigation offers the first evidence to support that the socioemotional distress of boys, especially among the ones who feel less gender typical, is related to school-level negative gender socialization by peers. Given the gender differences in the findings, we discuss our interpretations of the findings separately for boys and girls.

Boys and Gender Norm Salience

In schools in which lower gender typicality was associated with increased risk of victimization (i.e., higher gender norm salience), boys who felt that they were less gender typical in seventh grade reported greater loneliness and social anxiety in eighth grade. These effects, which were consistent with our hypotheses, were obtained even when taking into account socioemotional distress in the beginning of middle school (i.e., at sixth grade), personal experiences of peer victimization, and school-level general victimization. On the other hand, in environments that appeared to be more accepting of low gender typicality (i.e., lower gender norm salience), gender typicality was not associated with distress. Underscoring the effects of school gender climate, these findings extend research demonstrating that individuals’ experiences with lack of peer acceptance (T. E. Smith & Leaper, 2006) gender-based teasing (Jewell & Brown, 2014), and other forms of peer victimization (D. S. Smith & Juvonen, 2017) partly account for associations between gender typicality and socioemotional distress.

Our contextual moderation analyses demonstrated that for boys reporting low to average gender typicality, more salient gender norms were associated with greater loneliness. Additionally, higher gender norm salience appears to have detrimental effects on depressed mood for boys regardless of their gender typicality—even highly gender typical boys in such environments reported more depressed mood. These effects of norm salience are similar to the effects of felt pressure at the individual level, in that they are documented over and above the effects of gender typicality (Egan & Perry, 2001). This finding for depressed mood aligns with prior research demonstrating that school-level pressures surrounding particular norms can impact socioemotional distress—even for those who may not be victimized (e.g., Lampard et al., 2014). It is also consistent with evidence suggesting that pressure to conform to masculine gender roles is related to increased depression among men (e.g., Good et al., 1996). Even highly typical boys might suffer from increased depressive symptoms when facing pressure to conform, if, for example, they feel that they should not express sadness or seek help when facing any sort of difficulty because of the content of masculine gender norms (Good & Wood, 1995). Such pressures, e.g., to be tough and overtly heterosexual, are particularly pervasive during adolescence (e.g., Pascoe, 2012). Thus, school-based gender norm salience may be problematic not only for young adolescent boys feeling less gender typical, but for all boys inasmuch as such salience may reflect greater conformity pressure.

Differences in Girls’ Experiences

In contrast to boys, within-school gender norm salience was not associated with girls’ loneliness, social anxiety, and depressed mood over time. There was no support for the contextual moderator hypothesis, nor did gender norm salience alone predict girls’ distress. The lack of gender norm salience findings begs the question of why school social context does not appear to factor in for girls in this case. There are multiple potential explanations for these findings. For one, gender norm salience did not vary across schools as much for girls as it did for boys. Thus, limited between-school variability may have contributed to the lack of school-level differences in girls’ distress. Another possibility is that, while messages that boys receive about what it means to be a boy tend to be fairly uniform and center strongly and consistently around homophobia and toughness starting in childhood (Kane, 2006) and continuing into adolescence (Pascoe, 2012) and adulthood (Good & Wood, 1995), the messages girls receive about gender norms may be mixed (Eder et al., 1995). Non-gender-typical behavior (e.g., rough-housing) may be tolerated and even encouraged among girls, particularly when they are young (Kane, 2006). During adolescence, the picture becomes less clear; some stereotypically masculine traits are simply relabeled (e.g., tomboys may become jocks), while in other instances girls are pressured to be more stereotypically feminine (Carr, 2007). In short, both expectations for girls and reinforcement of these expectations may be far more variable than for boys, creating more noise in our index of gender norm salience. This possibility warrants further exploration with more in-depth measures of gender conformity, gender typicality, and peer gender socialization.

Yet another potential explanation is that for girls, a good amount of pressure surrounding gender norms may come from sources other than peers during adolescence, and at the same time the shift from childhood to adolescent gender expectations may be greater than for boys. On average, girls also begin puberty earlier than boys (Sun et al., 2002). As such, they may also experience any increases or shifts in gender norm expectations earlier. For example, in a study of women who identified as tomboys when they were children, a number of participants reported that they were pressured to act in a more “feminine” manner during adolescence, sometimes specifically because of physical changes, such as breast development (Carr, 2007). In addition, in the same study, it was parents, rather than peers, who were more often reported to exert the most pressure to conform. If young adolescent girls are facing substantial conformity pressure at home, this may in part explain both the lack of effect of norm salience (given that the index used addresses only peer socialization pressures) and the consistent association between gender typicality and socioemotional distress. In short, comparative lack of variability in level of gender norm salience, combined with possible increased variability in both sources and content of gender norm salience, may have obscured its effects on girls’ distress in comparison to boys’. Gender norms may be made more salient by earlier and more obvious physical development, parental and other societal pressures, etc. in addition to gender-related peer victimization, which would account for the consistent association between gender typicality and socioemotional distress regardless of school environment.

Contributions, Limitations, and Future Studies

The current study extends prior research in several ways. First, relatively little is known about differences between schools for any type of gender socialization by peers, and we are not aware of any studies that examine the effects of school differences in gender norm salience on loneliness, social anxiety, and depressed mood. Further, we expanded on cross-sectional analyses of school-level gender climate effects (e.g., Toomey et al., 2012) by capitalizing on longitudinal data across three years of middle school. Second, we used a novel statistical method to capture variation in such gender norm salience across schools. To ensure the salience index did not merely capture individual experiences of victimization or higher school-level rates of such mistreatment, we also controlled for victimization at the individual and school level. Finally, this investigation was conducted with youth recruited from 26 ethnically diverse middle schools. Although gender socialization and thus salience of gender norms may be particularly intense among some ethnic groups (e.g., Brown, Craig, & Halberstadt, 2015), we did not test gender norm salience effects separately for each ethnic group due to insufficient numbers of boys and girls of each ethnicity in some schools. However, in the current study we controlled for ethnic difference in the emotional outcomes, and the diversity of the sample increases the generalizability of our findings across demographic groups.

One way in which this study could be improved would be by utilizing multiple sources of data. Gender typicality and peer victimization (and as such, gender norm salience as well) were self-reported, as was socioemotional distress. By examining the association between self-reported gender typicality and self-reported peer victimization, we were able to capture aspects of adolescents’ subjective realities which are highly relevant when considering socioemotional outcomes (e.g., Juvonen et al., 2001). In addition, the finding that gender norm salience was related to boys’ depressed mood regardless of their gender typicality suggests that the current findings indeed capture an aspect of the schools’ climate rather than merely reflect the feelings of the boys who do not feel as typical of their gender. However, shared method variance may have contributed to the associations found, and therefore future studies using teacher- or peer-reports are needed. For example, it would be particularly interesting to examine whether peer perceptions of gender typicality, which were not available for this study, would function similarly to self-perceptions.

In addition, although the participants in the study were ethnically diverse, the sizes of ethnic groups in each school did not allow for thorough testing of differences in associations between gender typicality and distress based on ethnicity. It is plausible that peer gender socialization differs across ethnic groups, much like parent gender socialization (e.g., Brown et al., 2015). Culture-specific conceptions of gender, conflation of racial and gender stereotypes, and the highly limited available studies showing racial moderation of the association between gender identity and distress (e.g., Corby, Hodges, & Perry, 2007) suggest that more nuanced approaches are needed.

Further, the measure of gender norm salience used in the current study provides a novel way to address gender socialization at a group level, without requiring that participants are able to recognize and explicitly report on school-wide gender-related victimization. However, the current approach, using a measure of peer victimization which is not specific to gender-related mistreatment, does not provide details about the nature of such gender socialization and experienced or observed victimization, such as what types of deviations from gender norms are most likely to elicit negative reactions from peers. For example, is it particular mannerisms as opposed to appearance of boys that evoke most negative peer reactions from others boys? To complement the current school-level measure, a combination of observation and interview (e.g., Carr, 2007) and direct measures of gender typicality-related teasing experienced directly (e.g., Jewell & Brown, 2014), gender norm-related victimization observed in school (e.g., Kosciw et al., 2014), and individual perceptions of gender-related school climate (e.g., Toomey et al., 2012) would provide a more nuanced picture than can the current measure in isolation. In addition, experimental methods (e.g., Heinze & Horn, 2014) would allow more controlled examination of various types of violations of gender norms.

Conclusion

The current study provides new insight into the role of gender norm salience in the socioemotional distress of young adolescents. Examinations of school differences in peer socialization and salience of gender norms provide vital insights into the potential impact of young adolescents’ social context on their wellbeing. That is, patterns of targeted victimization are not only the social experience of particular youth, but can also capture the salience of particular sets of norms within a social collective (i.e., boys in a particular school). Understanding of such school differences can offer insights into fundamentally different intervention strategies—e.g., strategies that seek to reshape the broader peer culture—rather than approaches that consider low gender typicality as a personal liability. For example, when boys make gender norms more salient by victimizing other boys based on gender expression, such behavior may be seen as normative and met with a “boys will be boys” attitude (Kochenderfer-Ladd & Pelletier, 2008). Moreover, in some cases, punitive gender socialization is also perpetuated by teachers (e.g., Kosciw et al., 2014). Just as increased salience of gender norms due to peer victimization has negative implications even for gender typical boys, such adult participation or lack of intervention may have detrimental effects as well. School-based prevention and intervention should thus focus in part on encouraging adult authority figures to take such issues seriously, rather than ignoring it. Gender is rarely a primary focus of diversity trainings in teacher education programs (Jennings, 2007), and as such teachers may need to be educated on the detrimental effects of such behavior. In addition, there should be clear policies outlining how to appropriately address victimization associated with gender norms, and disciplinary structures in place for adults, as well as peers, who victimize gender atypical youth.

In addition to improving reactions to differing gender expression, schools might be able to reduce the salience of gender norms by taking steps to broaden perceptions of those norms and shift views of what it means to be gender typical. Although it would be difficult to address all of the possible factors contributing to perceived gender typicality, it is plausible that being strategic about the gender norm messages given at school could lessen feelings of atypicality without pushing children to conform to stereotypes. Along with careful consideration of any language reinforcing particular gender norms or stereotypes (e.g., boys will be boys, or dress more like a lady; Pascoe, 2012), curriculum and media could be modified to be more gender-inclusive. Instead of tokenism (i.e., occasionally highlighting the contributions of a female scientist or male dancer) which can be counterproductive (Brookfield, 2007), deliberately diversifying curriculum to include individuals who do not fit mainstream gender norms—whether this has to do with their appearance, sexuality, career, or mannerisms—is needed. In other words, if gender norm deviations are normalized and become familiar, perhaps they may cease to be perceived as deviations, thus reducing the salience of gender norms.

Biographies

Danielle Sayre Smith is a doctoral student at the University of California, Los Angeles. Their major research interests include identity, social group norms, and intergroup attitudes during adolescence, and their impact on peer relationships and adjustment.

Hannah L. Schacter is a doctoral student at the University of California, Los Angeles. Her major research interests include risk factors and consequences of adolescents’ experiences of peer victimization.

Dr. Craig Enders is a professor at the University of California, Los Angeles. His major research interests include multiple imputation, maximum likelihood and Bayesian estimation with incomplete data, multilevel modeling, and longitudinal modeling.

Dr. Jaana Juvonen is a professor at the University of California, Los Angeles. Her major research interests include peer relationships and their associations with development and adaptive functioning of youth, with foci on person-group mismatch and school context.